Deterministic core, AI at the edges
The boring, predictable layer does the work you trust with money. AI is summoned for exceptions the deterministic core has already flagged — never as the thing holding the pen.
About
Noctix is the software studio of Lucas Utreras Acevedo — a software engineer who builds production systems that run business operations autonomously.
Most recently that meant a billing platform trusted with six figures a month: it detects flights, reads manifests, reconciles weights, writes ledgers, and issues invoices twice a day without anyone at the keyboard. Designed, built, and operated end to end.
The name comes from nox — Latin for night. The real test of operational software is what happens while nobody is watching.
Good operational software should keep working after hours, verify its own output before it commits anything, and wake a human only when something genuinely needs one. Anything less isn't automation — it's a task you now have to supervise. That standard is the whole reason this studio exists, and it's the bar every Noctix system is built to.
Lucas works in English and Spanish, with clients and their teams in both.
How Noctix thinks
The boring, predictable layer does the work you trust with money. AI is summoned for exceptions the deterministic core has already flagged — never as the thing holding the pen.
Anything that writes to a ledger, an invoice, or a customer record re-checks itself against the source first. Halting is always better than being confidently wrong.
A healthy system should be quiet. Alerting is tiered so that noise means something, and a message in your inbox is a real signal rather than background hum.
That's precisely the kind of problem Noctix exists for.